Management of stage I and II nonsmall cell lung cancer

ERJ Express. Published on December 23, 2016 as doi: 10.1183/13993003.00764-2016 SERIES THORACIC ONCOLOGY Management of stage I and II nonsmall cell lung cancer Fiona McDonald 1,6, Michèle De Waele 2,6, Lizza E. L. Hendriks 3,6, Corinne Faivre-Finn 4,5, Anne-Marie C. Dingemans 3 and Paul E. Van Schil 2 Number 2 in the series Multidisciplinary questions in thoracic oncology: the team experience Edited by J-P. Sculier Affiliations: 1 Dept of Radiotherapy, Royal Marsden NHS Foundation Trust, Sutton, UK. 2 Dept of Thoracic and Vascular Surgery, Antwerp University Hospital, Antwerp, Belgium. 3 Dept of Respiratory Disease, Maastricht University Medical Centre, Maastricht, the Netherlands. 4 Manchester Academic Health Science Centre, Institute of Cancer Sciences, Manchester Cancer Research Centre (MCRC), University of Manchester, Manchester, UK. 5 Radiotherapy Related Research, Christie NHS Foundation Trust, Manchester, UK. 6 These authors equally contributed to this manuscript. Correspondence: Paul E. Van Schil, Dept of Thoracic and Vascular Surgery, Antwerp University Hospital, Wilrijkstraat 10, B-2650 Edegem, Antwerp, Belgium. E-mail: paul.van.schil@uza.be @ERSpublications Early stage lung cancer treatment needs multidisciplinary cooperation between physicians, oncologists and surgeons http://ow.ly/ekxe304xhmr Cite this article as: McDonald F, De Waele M, Hendriks LEL, et al. Management of stage I and II nonsmall cell lung cancer. Eur Respir J 2017; 49: 1600764 [https://doi.org/10.1183/13993003.00764-2016]. ABSTRACT The incidence of stage I and II nonsmall cell lung cancer is likely to increase with the ageing population and introduction of screening for high-risk individuals. Optimal management requires multidisciplinary collaboration. Local treatments include surgery and radiotherapy and these are currently combined with (neo)adjuvant chemotherapy in specific cases to improve long-term outcome. Targeted therapies and immunotherapy may also become important therapeutic modalities in this patient group. For resectable disease in patients with low cardiopulmonary risk, complete surgical resection with lobectomy remains the gold standard. Minimally invasive techniques, conservative and sublobar resections are suitable for a subset of patients. Data are emerging that radiotherapy, especially stereotactic body radiation therapy, is a valid alternative in compromised patients who are high-risk candidates for surgery. Whether this is also true for good surgical candidates remains to be evaluated in randomised trials. In specific subgroups adjuvant chemotherapy has been shown to prolong survival; however, patient selection remains important. Neoadjuvant chemotherapy may yield similar results as adjuvant chemotherapy. The role of targeted therapies and immunotherapy in early stage nonsmall cell lung cancer has not yet been determined and results of randomised trials are awaited. Received: April 15 2016 Accepted after revision: Oct 04 2016 Conflict of interest: Disclosures can be found alongside this article at erj.ersjournals.com Previous articles in this series. No. 1: Malhotra J, Malvezzi M, Negri E, et al. Risk factors for lung cancer worldwide. Eur Respir J 2016; 48: 889 902. Copyright ERS 2017 https://doi.org/10.1183/13993003.00764-2016 Eur Respir J 2017; 49: 1600764 Copyright 2016 by the European Respiratory Society.

Introduction Multimodality treatment of stage I and II nonsmall cell lung cancer (NSCLC) requires a delicate interplay between surgery, radiotherapy and chemotherapy. More recently, immunotherapy and targeted agents have emerged as potentially important treatment modalities. To determine management for each patient, a thorough knowledge is required of the natural history of disease, risk assessment of the individual patient, evaluation of the diagnostic and staging examinations available, multidisciplinary input into individualised treatment plans and, importantly, discussion of the risks and benefits of treatment options with the patient. To this end, every patient with presumed or proven lung cancer should be discussed within a dedicated multidisciplinary team (MDT) to ensure the optimal individualised therapeutic plan. In this article, as part of the series on Multidisciplinary questions in thoracic oncology, the different treatment modalities that are currently available for these patients are discussed in depth. The local modalities, surgery and radiotherapy, and their integration with systemic treatment are reviewed. Each co-author searched the literature over the past 10 years in their specific discipline. The main databases used were PubMed, Cochrane database and Web of Science. References of the selected papers and abstracts of major meetings were also screened for new, relevant data. As a broad area is covered, this manuscript cannot be considered a systematic review stricto sensu but a clear update is provided on management of stage I and II NSCLC. When applicable, levels of evidence are mentioned (table 1). Presently there are a number of grey areas, highlighting the need for further clinical research to provide high-level evidence for future recommendations. Diagnosis, patient selection and staging Computed tomography screening The incidence of early stage NSCLC is expected to increase due to wider availability of computed tomography scans and the introduction of screening in high-risk populations [1]. This will increase surgical workload. The higher risks of surgical morbidity and mortality in patients with multiple co-morbidities, coupled with an ageing population suggests that non-surgical management of early stage NSCLC is likely also to rise [2]. Implication of changes in World Health Organisation pathological classification The new World Health Organization (WHO) classification of lung tumours, published in 2015, integrates immunohistochemistry into the classification of resected lung cancers [3]. A complete histological evaluation of the tumour is necessary for diagnosis and is of prognostic value. In this edition, the American Thoracic Society (ATS)/European Respiratory Society (ERS)/International Association for the Study of Lung Cancer (IASLC) classification of adenocarcinoma published in 2011 was included without changes [4]. New subcategories include adenocarcinoma in situ and minimally invasive adenocarcinoma. Complete histological review should be performed to look for invasive foci with measurement of invasion. The term bronchioloalveolar carcinoma, which gave rise to much confusion, is not used anymore. Subtypes of invasive adenocarcinoma should be listed with their relative percentage indicated. This classification also has profound surgical implications which are discussed in the surgical section [5]. Implication of changes in 8th edition of tumour node metastasis classification In 2017, the 8th tumour node metastasis (TNM) classification of lung cancer will be introduced [6, 7]. Important changes have been made in the T descriptor on the basis that larger tumours are associated with worse prognosis. 3 cm diameter remains the cut-off margin between T1 and T2; T1 is subdivided into T1a ( 1 cm), T1b (1.1 2 cm) and T1c (2.1 3 cm). T2 is subdivided into T2a (3.1 4 cm) and T2b (4.1 5 cm). Tumours larger than 5 cm and 7 cm are categorised as T3 and T4, respectively. Using the entire database available for the 8th TNM classification, 5-year survival has improved (level C evidence). Overall survival by clinical and pathological stage for the 7th and 8th TNM classifications are given in tables 2 and 3, respectively, for comparison [6 8]. TABLE 1 Levels of evidence Level A B C D E Description Randomised controlled trials with a consistent pattern and rich body of data Randomised controlled trials with a limited number of patients or inconsistent results Non-randomised trials, observational studies Panel consensus judgment Expert opinion https://doi.org/10.1183/13993003.00764-2016 2

TABLE 2 Overall survival by stage according to the 7th tumour node metastasis (TNM) classification Stage 7th edition 5-year survival for clinical stage 5-year survival for pathological stage IA 82% 83% IB 66% 71% IIA 52% 57% IIB 47% 49% 7th TNM classification based on [8] Cardiopulmonary risk assessment All patients should be discussed in a MDT to decide which treatment option is most suitable on an individual basis. Although there is no clear definition of high-risk patients, it should be kept in mind that empiric selection of patients may deny optimal oncological management [9]. Even with modern computed tomographic and positron emission tomographic (PET) staging, pathological upstaging is seen in up to 30% of patients [10, 11]. Multiple guidelines exist on selecting patients who are fit for surgery [12 14]. The second European Society for Medical Oncology (ESMO) consensus conference on lung cancer recommended detailed assessment of cardiac and pulmonary function in order to estimate surgical morbidity (level C evidence). Cardiac function is tested using the recalibrated thoracic revised cardiac risk index, made up of history of ischaemic heart disease and cerebrovascular disease, serum creatinine and whether the intended procedure is a pneumonectomy or not. Pulmonary function is determined by the percentage of predicted forced expiratory volume in 1 s (FEV1) and lung diffusion capacity (DLCO). When percentages are <80% for either one, exercise testing is recommended (maximal oxygen consumption and split lung function) (level C evidence) [15]. Also, for patients undergoing radiotherapy and chemotherapy, cardiopulmonary function should be assessed. This is especially important in patients with low DLCO and poor performance status to decide whether they tolerate the anticipated treatment scheme or not. Lung cancer staging For a presumable clinical stage I or II NSCLC evaluation of locoregional lymph node spread and screening for distant metastases are required. This is mostly performed by PET computed tomography scanning which provides anatomical and metabolic information. In case of suspicious lymph nodes or distant lesions, every effort should be made to obtain a pathological diagnosis by a minimally invasive technique (MIT) or invasive technique. Further lymph node evaluation is performed by endobronchial ultrasound or endoscopic ultrasound. In case results are negative, mediastinoscopy or thoracoscopy are the next steps (level D evidence). Recently, the European Society of Thoracic Surgeons (ESTS) updated guidelines for preoperative staging of NSCLC [16]. Also, for patients undergoing stereotactic body radiotherapy (SBRT) lymph node staging is recommended [17]. Treatment modalities Surgery Introduction Surgical intervention remains the gold standard for the approximately 30% of NSCLC patients who present with resectable stage I and II disease [18] and who are functionally operable (level C evidence) [5, 15, 19]. TABLE 3 Overall survival by stage according to the 8th tumour node metastasis (TNM) classification Stage 8th edition 5-year survival for clinical stage 5-year survival for pathological stage IA1 92% 90% IA2 83% 85% IA3 77% 80% IB 68% 73% IIA 60% 65% IIB 53% 56% 8th TNM classification based on [6, 7] https://doi.org/10.1183/13993003.00764-2016 3

The extent of resection and precise surgical approach are the subject of discussion. Additionally, with the advances in newer ablative techniques, their role in early stage disease is currently debated, especially in medically compromised patients. In this section, the indications for sublobar resections (SLR), sleeve resections and MIT by video-assisted thoracic surgery (VATS) and robot-assisted thoracic surgery (RATS) are discussed. Conservative interventions: SLR and bronchoplastic procedures Lobectomy has remained the standard of care for resection of early stage NSCLC since the prospective randomised Lung Cancer Study Group trial, comparing lobectomy with SLR (anatomical segmentectomy or wedge resection) for stage I NSCLC, which was published in 1995 [20]. The limited resection group had a three-fold increased incidence in local recurrence ( p=0.008), a 30% increase in overall death rate ( p=0.08) and a 50% increase in cancer-related death ( p=0.09) compared with patients undergoing lobectomy (level B evidence). It is important to note that the limited resection arm included anatomical segmentectomies as well as non-anatomical wedge resections, and tumours up to 3 cm were eligible. Since this study, there has not been another published prospective trial on this topic. Several single centre retrospective studies have been published with conflicting conclusions [21, 22]. LANDRENEAU et al. [23] performed a propensity-matched comparison between anatomical segmentectomy versus lobectomy for clinical stage I NSCLC. The study reported no significant difference in locoregional recurrence ( p=1.00) or in 5-year disease-free survival (DFS) ( p=0.47). YANO et al. [24] published a multi-institutional retrospective study (n=1737) on limited resections (segmentectomy or wedge resection) for ct1 N0 M0 NSCLC. Computed tomography imaging was used to determine the invasive potential of the tumour on the basis of the ratio of consolidation (C) to the maximal tumour diameter (T) (C/T ratio). Tumours were classified invasive if the C/T ratio was >0.25 and non-invasive if C/T 0.25. Overall survival and DFS after limited resection were 94.0% and 91.1% at 5 years, respectively. C/T 0.25 predicted for good outcome, especially in ct1a N0 M0 disease. In a meta-analysis by CAO et al. [25] data were reviewed of intentional SLRs versus lobectomy for early stage NSCLC. Patients who underwent SLR for small, peripheral NSCLC after intentional selection rather than ineligibility for larger resections, achieved similar long-term survival outcomes as those who underwent lobectomies (level C evidence). However, patients included in this meta-analysis were a highly selected cohort and these results should be interpreted with caution. Two large prospective trials are currently ongoing (CALGB 140503 and JCOG0802/WJOG4607L), comparing segmentectomy versus lobectomy in NSCLC and the results are eagerly awaited (table 4) [26, 27]. SLR is a valid alternative to lobectomy in lung cancer patients who meet the following criteria: stage IA disease, tumours up to 2 cm in diameter, peripheral tumour location and predominantly ground-glass (non-solid) appearance on computed tomography imaging (level C evidence) [19]. Anatomical segmentectomy is preferred over wedge resection since the latter is associated with higher rates of locoregional recurrence in stage IA NSCLC [28, 29]. Although, in case of a predominant ground glass opacity clinical stage IA adenocarcinoma, a wedge resection might be performed with a T1a tumour, and segmentectomy for a T1b tumour due to the low-grade malignancy and favourable prognosis (level C evidence) [30]. When dealing with a solid-type, clinical stage IA NSCLC, a lobectomy is recommended [31]. TABLE 4 Current randomised controlled trials comparing lobectomy versus sublobar resection US CALGB 140503 trial (NCT00499330) [26] Japanese trial (JCOG0802/WJOG4607L) [27] Surgical study arm 1 Lobectomy Lobectomy Surgical study arm 2 Limited resection (anatomical segmentectomy Anatomical segmentectomy or non-anatomical wedge) Eligibility Peripheral lung nodule 2 cm on perioperative CT scan and located in outer third of lung Peripheral NSCLC (suspicion) 2 cm (proportion max. diameter of tumour to consolidation >0.5) and presumed to be NSCLC Primary end-point DFS OS Secondary end-points OS, LR, systemic recurrence rates, PFT, correlation preoperative PET-CT with outcome, false negative rate for PET if nodal metastases; utility of annual follow-up CT Target recruitment 1297 patients 1100 patients PFT, relapse-free survival, LR, proportion of completion of segmentectomy, LOS, duration chest tube, operation time, blood loss, number of auto-sutures CT: computed tomography; NSCLC: nonsmall cell lung cancer; DFS: disease-free survival; OS: overall survival; LR: local recurrence; PFT: pulmonary function test; PET: positron emission tomography; LOS: length of stay in hospital. https://doi.org/10.1183/13993003.00764-2016 4

The second ESMO consensus conference on lung cancer concluded that SLR is acceptable for pure ground glass opacities and adenocarcinoma in situ with minimal invasion [15]. Lobectomy is still the preferred treatment for tumours 2 cm with solid appearance on computed tomography (level C evidence). Either open thoracotomy or VATS can be utilised as per experience of the thoracic surgeon. Next to the aforementioned distal lung parenchyma saving procedures, proximal bronchoplastic interventions as sleeve resections may be the treatment of choice in early stage NSCLC with proximal bronchial involvement or positive N1 nodes around a lobar bronchus [32, 33]. In this way, a pneumonectomy is avoided. Conservative resections are more often performed in carcinoids or in patients with impaired pulmonary reserve [34]. Minimally invasive surgery MIT like VATS or RATS have been widely implemented as standard treatment for early stage NSCLC. Both procedures are equivalent in outcome (level C evidence) [35]. A retrospective analysis on the National Cancer Database (USA) looked at perioperative outcomes and survival of patients with clinical T1 2, N0 M0 NSCLC undergoing open or MIT: VATS and RATS [36]. This is an oncology database established by the American College of Surgeons and the American Cancer Society with data from hospital registries that are collected in more than 1500 Commission on Cancer-accredited facilities. Shorter median length of stay (5 versus 4 days; p<0.001), and improved 2-year survival (87% versus 86%; p=0.04) were observed when an MIT was used. There was no significant difference in nodal upstaging rates and 30-day mortality between the two groups. Comparing both MITs, there was no significant difference between VATS and RATS in regards to nodal upstaging, 30-day mortality and 2-year survival rates. It is important to consider the role of nodal dissection, for which precise criteria have been established, to accurately stage the extent of cancer spread pathologically [37]. MEDBERY et al. [38] conducted a retrospective analysis on the same National Cancer Database and reported nodal upstaging was more frequent in patients treated with lobectomy by thoracotomy than by VATS (12.8% versus 10.3%; p<0.001). This difference was non-significant in patient groups treated in academic research facilities [38]. Differences in quality of life measures following open compared with VATS anatomical resection were assessed in a prospective study and were found to be similar in both patient groups (level C evidence) [39]. In addition the patient-reported physical component summary and pain scores after thoracotomy and VATS were also similar in both groups during the first 12 months after surgery [39]. Radiotherapy Introduction In addition to the advances in surgery over recent years, there have been dramatic developments in radiotherapy for NSCLC patients. Radiotherapy is established as an alternative curative treatment option for patients with early stage disease, particularly in patients who are considered medically inoperable due to co-morbidities. In this section, we compare local ablative therapies for stage I NSCLC and discuss the role of radiotherapy in a multidisciplinary setting as adjuvant and definitive treatment for stage I and II NSCLC. Comparison of local ablative treatments for stage I NSCLC Whenever there is a contraindication for surgery, radiotherapy may offer a valid alternative. To date, no valid prospective randomised controlled trial (RCT), comparing surgery and SBRT in medically operable patients with early stage NSCLC, has been completed; therefore, the question of whether SBRT yields similar results as complete surgical resection, remains unanswered (level C evidence) [40]. A recently performed propensity matched comparison between surgery and SBRT in over 117 000 patients with stage I NSCLC, derived from the National Cancer Database demonstrated that, although 95% of patients received surgery, the median overall survival favoured the surgical group (62.3 versus 33.1 months; p<0.001) [40]. The main limitation of such analysis is the difficulty in attributing improved outcome to patient selection or better cancer control, particularly since causes of death are unknown. Unfortunately, prospective RCTs, comparing surgery and SBRT in medically operable patients with early stage NSCLC, have all been terminated prematurely due to poor accrual (table 5). A recent combined analysis of patients randomised in both the Dutch ROSEL trial and the US STARS trial comparing SBRT with lobectomy demonstrated a significant 3-year overall survival advantage in favour of SBRT (95% versus 79%; p=0.037) in the 58 included patients [41]. DFS was similar in both groups; however, severe grade 3 or more toxicity was lower in the SBRT group (10% versus 44%). Despite small numbers the data suggest at least clinical equipoise between the two treatment modalities (level B evidence). However, as discussed in several letters to the Editor, it should be noted that mortality in the surgical arm was unacceptably high, that histology was not obtained in every case and that direct comparison of locoregional control between surgery and radiotherapy is not possible [43 49]. Further interest lies in comparing SBRT to surgical resection (lobectomy or SLR) in patients considered https://doi.org/10.1183/13993003.00764-2016 5

TABLE 5 Completed randomised controlled trials comparing SBRT versus surgery Dutch ROSEL trial (NCT00687986) [41] US STARS trial (NCT00840749) [41] US ACOSOG z4099 trial (NCT01336894) [42] Surgical study arm Anatomical resection with lymph node Anatomical resection SLR ± brachytherapy dissection SBRT study arm 60 Gy in three fractions or 60 Gy in five 60 Gy in three fractions or 60 Gy in Variable dose in three fractions fractions four fractions Eligibility Operable patients, stage IA disease Operable patients, stage IA & IB disease 4cm High operable risk patients, stage IA disease Primary end-point 5 year local control 3 year OS 5 year OS Secondary end-points Toxicity, OS, quality adjusted life years Toxicity, progression-free, Toxicity, DFS disease-specific survival Recruitment 22 patients 36 patients 13 patients SBRT: stereotactic body radiotherapy; OS: overall survival; SLR: sub-lobar resection; DFS: disease-free survival. high-risk for surgery. There are ongoing studies, assessing randomisation in this setting, currently recruiting (table 6). The use of radiofrequency ablation (RFA) might rival with SBRT or SLR in early stage NSCLC. One study has compared the selection criteria and short-term outcomes in three prospective clinical trials that used SBRT (Radiation Therapy Oncology Group (RTOG) trial 0236), SLR (American College of Surgeons Oncology Group (ACOSOG) trial Z4032) and RFA (ACOSOG trial Z4033) [50]. The overall 90-day mortality for SBRT, surgery and RFA was 0%, 2.4% and 2.0%, respectively ( p=0.5) (level C evidence). The RFA trial included older patients with more impaired lung function. Another study has assessed the outcomes of SLR, RFA and radiation treatment in 116 patients with histologically proven stage I NSCLC from a prospective database [51]. The hazard ratio for primary tumour recurrence adjusted for age and tumour size was 2.73 (95% CI 0.72 10.27) for SLR versus radiotherapy and 7.57 (95% CI 1.94 29.47) for SLR versus RFA. SLR was associated with a higher primary tumour control rate compared with RFA or radiotherapy but no differences were observed in overall survival or DFS (level C evidence). Interpretation of data from both these studies is limited as baseline patient characteristics were not comparable and treatment was not randomly assigned. In particular in the second study the radiotherapy group included patients treated with both SBRT (57%) and conventionally fractionated radiotherapy (43%). The median tumour size was significantly larger in the radiotherapy group compared with the other two groups. A large prospective RCT is required to assess the benefits of RFA in comparison with SBRT and surgery. Adjuvant radiotherapy in resected early stage NSCLC A meta-analysis of data from nine randomised trials of adjuvant radiotherapy after resection of stage I III NSCLC revealed a significant absolute detriment of 7% increased mortality at 2 years with the addition of TABLE 6 Planned/ongoing randomised controlled trials comparing SBRT versus surgery US POSTLIV trial (NCT01753414) US STABLE-MATES trial (NCT02468024) US Trial Mayo Clinic (NCT01622621) UK SABRTooth trial (NCT02629458) Surgical study arm Complete resection SLR SLR Complete resection and lymph node dissection SBRT study arm 55 Gy in five fractions 54 Gy in three fractions 54 Gy in three fractions Various dose/fraction schedules Eligibility Lobectomy or pneumonectomy, tumour 3cm 2 year loco-regional control Time to local/distant failure, DFS, OS High operative risk tumour 4 cm High operative risk peripheral tumour 5 cm High operative risk peripheral tumour 5 cm Primary end-point 3 year OS 2 year OS Average recruitment 3 patients per month Secondary end-points Toxicity, progression-free Toxicity, progression-free survival survival Target recruitment 76 patients 258 patients 96 patients 54 patients SBRT: stereotactic body radiotherapy; SLR: sublobar resection; OS: overall survival; DFS: disease-free survival. https://doi.org/10.1183/13993003.00764-2016 6

radiotherapy (level A evidence) [52]. Subgroup analyses suggested that the adverse effect was greatest for patients with stage I and II disease and therefore adjuvant radiotherapy is not recommended for completely resected early stage disease [15, 53]. Definitive radiotherapy for stage I NSCLC SBRT makes use of advanced radiotherapy planning and delivery techniques to permit high doses of radiation per fraction to be given accurately to small discrete targets with high conformality and rapid dose fall-off within the surrounding normal tissues. The majority of published studies of lung SBRT are single-centre retrospective series (level C evidence). There are technical variations in practice in the literature including whether the reported radiation dose is prescribed to the 100% or to the isodose covering the target, in the type of planning algorithms software and whether a heterogeneity correction was applied. Additionally, there are variations in patient selection, including whether there was pathological confirmation of the treated lesion, staging, including whether the mediastinum was invasively staged and in the computed tomography scanning follow-up intervals for assessment of local disease control. Bearing these limitations in mind, promising outcomes are observed with SBRT; for example, a systematic review of published studies of SBRT for peripherally positioned stage I NSCLC reports 2-year local disease control rates of 91% and 2-year overall survival rates of 70% (level C evidence) [54]. Importantly, it is the position of the steep dose fall off within normal tissues that determines the potential toxicity. While toxicity with lung SBRT is generally low with grade 3 or higher rates usually <4% [55 57], the exception is for centrally placed lesions as highlighted in a single centre phase II study. Patients with centrally located tumours were treated with a schedule of 60 66 Gy in three fractions (54 Gy in three fractions equivalent with heterogeneity correction) and after 4 years follow-up had an 11-fold higher risk of developing grade 3 5 toxicities when compared with patients with peripherally located tumours [58]. This led to the initial definition of the no fly zone (the volume encompassing 2 cm in all directions around the proximal bronchial tree) used in the RTOG 0236 trial [57] and subsequent SBRT studies (level C evidence). Various dose fractionation schedules are reported in the literature varying between one and 10 fractions with higher number of fractions predominantly being used for more centrally located lesions; however, the optimal dose remains unknown and is complicated by the variation in dose prescription methods used in the literature, making comparison of outcomes challenging (level C evidence). While schedules with a biologically equivalent dose (BED) of 100 Gy10 or more are associated with high local disease control rates [59, 60], a meta-analysis suggests the highest biologically equivalent schedules (>146 Gy10) may be associated with lower survival rates than the medium-high schedules (106 146 Gy10) (level C evidence) [61]. To address the dose question for peripheral tumours, the randomised phase II RTOG 0915 trial compared 48 Gy in four fractions (BED 105 Gy10) to a single 34 Gy fraction (BED 150 Gy10) [62]. One-year overall survival was 85% in the single-fraction arm and 91% in the four-fraction arm with similar toxicity and local tumour control rates (level B evidence). Further follow-up is awaited to decide whether the single-fraction schedule will be compared to the RTOG 0236 schedule of 54 Gy in three fractions in a phase III setting for peripheral lesions. For central lesions, the initial results from the RTOG 0813 study investigating maximum tolerated dose for central lesions using five fractions, presented in abstract form, suggest that the highest investigated dose level of 60 Gy in five fractions is associated with a 7.2% risk of severe dose limiting toxicity [63]. Data on efficacy of the various dose levels in the phase II component of the study are awaited prior to decision about the dose to be considered in a phase III setting [63]. The European Organisation for Research and Treatment of Cancer (EORTC) LungTech trial (NCT01795521) investigating the safety of 60 Gy in eight fractions for centrally lesions is actively recruiting. When comparing studies with treatment of central lesions it is also important to note that there is more than one definition of a central lesion in the literature [64]. For inoperable patients, in comparison with standard fractionation radiotherapy, the Scandinavian phase II SPACE RCT compared 3D conformal radiotherapy with 70 Gy in 7 weeks with SBRT to 66 Gy in three fractions in inoperable patients with stage I peripheral lesions. Initial results, in abstract form, with a median follow up of 37 months reveal no significant difference in DFS or overall survival (level B evidence) [65]. However, the patients in the SBRT arm experienced significantly less any-grade pneumonitis (19% versus 36%) and oesophagitis (8% versus 30%) with improved dyspnoea, cough and chest pain quality-of-life measures. The phase III RTOG 0902 CHISEL trial (NCT01014130) completed recruitment and the results are awaited. There has been no direct comparison between SBRT and RFA in inoperable patients with early stage NSCLC; however, due to the greater body of evidence in the literature to support SBRT as definitive treatment for NSCLC, SBRT is considered the most appropriate therapy for inoperable patients with stage I NSCLC (level C evidence). Despite SBRT being well tolerated, patient selection remains important. The majority of published SBRT trials included lesions up to 5 cm in diameter. Larger lesions are included in multiple retrospective series https://doi.org/10.1183/13993003.00764-2016 7

and can be considered for SBRT if dose constraints to surrounding normal tissues can be met. Alternative more protracted schedules can also be considered, for example 60 Gy in 15 fractions [66] or 70 Gy in 17 fractions [67]. As for medically inoperable patients with poor lung function and multiple co-morbidities and advanced age, SBRT needs to be weighed up against the risks of no treatment. The median survival of patients with routinely detected clinical stage I and II disease is approximately 10 months [68], and population-based studies of elderly patients with early stage NSCLC suggest that the observed improvement in overall survival over time is limited to patients treated with radiotherapy, including SBRT, rather than surgery or to those not treated radically (level C evidence) [69]. Therefore, advanced age alone should not exclude patients from SBRT. With co-morbidities, there are no absolute contraindications to SBRT and, in general, patients with ECOG performance status of 0 2 or poor lung function [70, 71] should be considered for treatment. An important relative contraindication to SBRT, however, is active interstitial lung disease with higher than expected reported cases of severe or fatal pneumonitis in retrospective series [72, 73]. Up to 20% of patients treated with SBRT will relapse with distant metastases after treatment [74, 75]. There is no proven role for adjuvant chemotherapy following SBRT particularly in patients with larger lesions ( 4 cm) that would be considered for adjuvant therapy following resection. A study in this patient population would be interesting. However, given the majority of these patients are considered medically inoperable, many may not be suitable candidates for platinum-based systemic therapy. In summary, SBRT is the standard of care for medically inoperable patients with early stage peripheral NSCLC [15, 16]. In comparison with surgery in operable patients, SBRT remains an alternative and offers at least clinical equipoise with surgery, especially in those considered high risk for surgery until further evidence is available comparing the two modalities. Prospective trial data are awaited to determine the optimal dose schedules, in particular for central lesions [15, 76]. Definitive (chemo)radiotherapy for stage II NSCLC For the relatively small proportion of patients with medically inoperable stage II NSCLC not suitable for SBRT, usually because of ipsilateral hilar nodal involvement, the standard of care is treatment with chemo-radiotherapy (CRT) using conventional fractionation and platinum-based regimens [77 79]. The role of concomitant compared with sequential chemotherapy in this patient group is less clear. In the landmark meta-analysis of over 1200 patients with stage I III disease, concomitant CRT was associated with a 4.5% benefit in overall survival at 5 years compared with a sequential approach at the expense of a significant increase in acute oesophageal toxicity from 4% to 18% [80]. However, less than 3% of the patients included in the analysis had stage I II with the vast majority having stage III disease (level B evidence). The added benefit of accelerated radiotherapy in the sequential setting is also less clear in stage II disease (level C evidence). The large meta-analysis assessing accelerated hypo- or hyper-fractionated radiotherapy schedules compared to conventionally fractionated treatment in over 2000 patients with NSCLC [81] demonstrated that accelerated schedules were associated with a 2.5% improvement in overall survival at 5 years. However, again only a small proportion of included patients, <20%, had stage I II disease (level C evidence). Given the paucity of data on definitive CRT specifically in inoperable patients with stage II disease, a population-based outcomes study (>550 patients) was recently performed in this patient group [82]. Stage II NSCLC patients treated with concomitant or sequential CRT were included and a median overall survival of 20.5 months was found. This figure approximates to survival figures in stage III disease from historic phase III trials (level C evidence). There is likely to be adverse selection bias for these patients with stage II disease given they were considered medically inoperable. In summary, patients with inoperable stage II disease should be treated with definitive radiotherapy and consideration of the addition of chemotherapy concomitantly or sequentially should be given based on fitness to tolerate treatment. Further studies are required to assess the benefits of CRT in this population specifically. Additionally, the role of treatment dose intensification with isotoxic radiotherapy schedules [83], oncogene targeted systemic therapies, DNA damage repair and immune checkpoint inhibitors needs to be explored. Salvage surgery after stereotactic radiotherapy For recurrent or persistent NSCLC after SBRT, salvage surgery is a valid therapeutic option when a complete resection is feasible and cardiopulmonary functional assessment shows no contra-indication for the anticipated resection. Although surgical salvage is a relatively new concept in thoracic surgery, recent data show that it is feasible in selected patients, not only for NSCLC but also for lung metastases (level E evidence) [49, 84]. Technical difficulties in performing the resection are limited on the condition that the hilum and mediastinum are not irradiated beforehand. More long-term data are needed to determine its role more precisely. https://doi.org/10.1183/13993003.00764-2016 8

Systemic therapy Introduction In contrast to the previous sections (locoregional treatment), in this section systemic therapy is reviewed. Platinum-based chemotherapy, targeted agents, immunotherapy and anti-angiogenic agents as adjuvant therapies for resected early stage NSCLC are discussed. Adjuvant platinum-based chemotherapy The RCTs of adjuvant cisplatin-based chemotherapy versus observation in patients with resected stage I III NSCLC demonstrate statistically significant benefit for the addition of systemic therapy (level A evidence) [85 89]. An overview of these trials is provided in table 7. A meta-analysis of pooled data (n=4584) revealed a 5.3% (overall survival) and 5.2% (DFS) improvement at 5 years with the addition of cisplatin-based chemotherapy [90]. However, a statistically significant interaction between disease stage and chemotherapy effect was observed. In stage IA, a potential detrimental effect was found with the addition of chemotherapy. In stage IB, a trend was seen favouring the addition of systemic therapy (level B evidence). A subsequent study of adjuvant carboplatin/paclitaxel in patients with resected stage IB NSCLC demonstrated no overall survival benefit; however, an unplanned retrospective subgroup analysis revealed a benefit for patients with tumours 4 cm [91]. Therefore, the ESMO guidelines state that adjuvant chemotherapy can be considered for the latter [15]. As mentioned previously, in the forthcoming eighth TNM edition, it is proposed to reclassify tumours 4 cm as T2b [6] and, subsequently, these are grouped as stage IIA instead of stage IB [7]. Recently, a Cochrane meta-analysis on adjuvant chemotherapy (without radiotherapy) in resected NSCLC was performed: 35 trials were identified and an analysis based on 8447 participants showed a clear benefit of adjuvant chemotherapy (HR 0.86, 95% CI 0.81 0.92; p<0.0001) with a 5-year absolute survival benefit of 4% (level A evidence) [92]. The most studied adjuvant drug combination is cisplatin/vinorelbine, although the Cochrane meta-analysis showed little variation in effect between different regimens (level A evidence). Moreover, an exploratory analysis of the E1505 trial (adjuvant chemotherapy±bevacizumab in early stage resected NSCLC) presented at the American Society of Clinical Oncology (ASCO) 2016 annual conference showed no difference in DFS or overall survival for the four different cisplatin-based chemotherapy regimens (cisplatin with investigator s choice: vinorelbine, docetaxel, gemcitabin or pemetrexed) [93]. No difference in survival has been demonstrated between neoadjuvant and adjuvant treatment (level B evidence) [94, 95]. Due to adjuvant trial results, most neoadjuvant trials were closed prematurely [95]. A potential advantage of neoadjuvant chemotherapy is that compliance may be better. Due to their clinical condition, more patients are likely to complete four cycles of neoadjuvant chemotherapy, compared with the 50 70% completing four treatment cycles demonstrated in most adjuvant studies [95, 97]. TABLE 7 Completed randomised, placebo controlled trials on (neo) adjuvant chemotherapy Chemotherapy ALPI trial [87] Big Lung Trial [88] IALT trial [85] JBR.10 [89] ANITA [86] Adjuvant MVP, three cycles every 3 weeks (Neo)adjuvant MIC, MVP, CV or NP, all three cycles every 3 weeks Adjuvant CV, NP, CE or C + vinblastin, three four cycles every 3 or 4 weeks Adjuvant NP four cycles every four weeks Adjuvant NP four cycles every four weeks Cisplatin dose 100 mg m 2 day 1 MIC/NVP: 50 mg m 2 80 120 mg m 2 50 mg m 2 days 1 100 mg m 2 day 1 day 1 CV/NP 80 mg m 2 day 1 and 8 Eligibility (stage) I, II, IIIA I, II, IIIA I, II, III IB, II IB, II, IIIA Primary end-point OS OS OS OS OS Secondary end-points DFS, toxicity DFS DFS, second primary DFS, toxicity, DFS, safety cancer, AEs safety Subjects n 1209 381 1867 482 840 Subgroups Elderly No age limit No age limit Upper age limit 75 years No age limit No age limit Co-morbidity Not specified Not specified fit to Not specified Only fit patients Not specified receive chemotherapy WHO performance status Not specified Not specified Not specified Only performance status 0 1 Performance status 0 2 WHO: World Health Organization; MVP: mitomycin, vindesine, cisplatin; MIC: cisplatin, mitomycin, ifosfamide; CV: cisplatin, vindesine; NP: cisplatin, vinorelbin; CE: cisplatin, etoposide; C: cisplatin; OS: overall survival; DFS: disease-free survival; AE: adverse events. https://doi.org/10.1183/13993003.00764-2016 9

Patients enrolled in clinical trials are highly selected and often younger, with good performance status and fewer co-morbidities compared with patients in the general population. In a retrospective analysis of the Ontario Cancer Registry, it was shown that adjuvant chemotherapy had a detrimental effect in patients with severe co-morbidities (Charlson score 3+) [98]. A subgroup analysis of the JBR 10 study demonstrated that elderly (>65 years) had the same amount of benefit with adjuvant chemotherapy than younger patients, without additional toxicity [99]. In the Lung Adjuvant Cisplatin Evaluation (LACE) meta-analysis the same was true for patients aged >70 years [100]. Interestingly, in both studies the elderly received on average fewer chemotherapy cycles and reduced cisplatin dose. Improved overall survival was noted for elderly (age >65 years) with stage I (tumour 4 cm) in a Surveillance, Epidemiology and End Results (SEER) -Medicare analysis (n=3289) although adjuvant chemotherapy was associated with an increased number of serious adverse events [101]. However, data on patients aged >75 years are lacking as this group was under-represented in clinical trials (level B evidence). In clinical trials, adjuvant chemotherapy is started within 6 8 weeks of surgery. In daily practice, 35% of patients start adjuvant chemotherapy >10 weeks after surgery; however, this did not appear to have a negative impact on survival [98]. The meta-analysis showed a clear overall survival benefit for adjuvant chemotherapy and it is advised in all current guidelines [100]. However, tools to optimally select patients who benefit from chemotherapy are warranted, especially since the updated survival analysis of the International Adjuvant Lung Cancer Trial (IALT) showed that the survival benefit did not persist after 5 years of follow-up, mainly due to increased non-lung cancer mortality in patients treated with adjuvant chemotherapy (level B evidence) [102]. The histological NSCLC subtype has not been shown to be a predictive factor of benefit from adjuvant chemotherapy and, as yet, there is no fully validated biomarker to identify patient subgroups who may derive particular benefit [92]. The IALT biomarker group has studied predictive value of several biomarkers including excision repair cross-complementation group 1 (ERCC1) and P53. ERCC1 expression seemed to be a predictive marker for response to platinum-based chemotherapy, with only ERCC1-negative tumours benefitting from adjuvant chemotherapy [103]. Although the randomised adjuvant Tailored Postsurgical Therapy in Early Stage NSCLC (TASTE) trial showed that a biology-driven randomised adjuvant trial is feasible, the planned phase III trial was terminated due to inaccuracy of the ERCC immunohistochemical staining classification [104]. Until now, no potential predictive biomarker has been validated in an RCT and none can be used to select patients who benefit from adjuvant chemotherapy. Invasive components of the tumour (vascular, lymphatic or perineural invasion) are also of prognostic significance in early stage NSCLC. Although not evaluated in RCTs, it is possible that resected early stage patients with adverse prognostic factors would benefit more from adjuvant treatment than those without. In a meta-analysis (22 studies, a total of 25 280 patients with resected stage I NSCLC), visceral pleural invasion was associated with death (HR 1.427, 95% CI 1.221 1.669; p<0.001) and recurrence (HR 1.600, 95% CI 1.284 1.995; p<0.001) [105]. This increased risk was found for all subgroups including patients with tumours <2 cm. Comparable results were found in another analysis including 13 cohort studies (27 171 patients) (level C evidence) [106]. Lymphovascular invasion (LVI) was also associated with a worse prognosis in a recent meta-analysis including resected stage I patients (20 studies, 8032 patients) [107]. Risk of death was significantly higher in patients with LVI (HR 1.81, 95% CI 1.53 2.14) (level C evidence). Conflicting studies exist on the prognostic significance of perineural invasion [108, 109]. A newer method to evaluate prognosis is the use of an immunoscore. For example, it has been shown that stromal CD8 + tumour-infiltrating lymphocytes density had a prognostic impact across all stages in multivariate analysis in a study including 797 resected stage I IIIA NSCLC patients [110]. 5-year overall survival was 61%, 50% and 41% for CD8 + high, intermediate and low scores, respectively (p<0.001) (level C evidence). Adjuvant targeted agents The effect of adjuvant epidermal growth factor receptor (EGFR) tyrosine kinase inhibitor (TKI) was tested in the BR19 study [111]. Unselected patients with completely resected stage IB-IIIA NSCLC were randomised between gefitinib or placebo for 2 years. This study was closed prematurely; however, it showed no benefit for adjuvant gefitinib (level B evidence) [111]. In addition, no benefit was shown in the subgroup with an activating EGFR mutation (n=15). With the proven efficacy of EGFR-TKIs in mutation-positive advanced disease, the role of adjuvant erlotinib versus placebo has been explored in patients with completely resected EGFR-expressing (immunohistochemistry) or EGFR-amplified (fluorescence in situ hybrisation) stage IB to IIIA disease [112]. Adjuvant erlotinib for 2 years did not prolong DFS in the EGFR mutation-positive NSCLC. Although there was a trend to improved DFS with gefitinib in the EGFR-activating mutation positive subgroup (n=161, HR 0.61 95% CI 0.38 0.98; p=0.039), this was not statistically significant due to hierarchical testing (level B evidence). Discussion points are the dose and duration of treatment. It might be that, in line with results in breast cancer, the treatment period was too short. Currently, there is no role for TKIs in the adjuvant setting, but further research in the EGFR mutant positive patients is warranted and trials are ongoing (table 8). https://doi.org/10.1183/13993003.00764-2016 10

https://doi.org/10.1183/13993003.00764-2016 11 TABLE 8 Planned/ongoing phase III randomised controlled trials comparing adjuvant EGFR-TKI to placebo after complete resection in EGFR mutant (exon 19 deletion/exon 21 L858R) NSCLC patients Experimental arm ADJUVANT (NCT 01405079) Gefitinib 250 mg once daily, 24 months International ADAURA trial (NCT 02511106) Osimertinib 80/ 40 mg once daily Comparator arm NP four cycles every three weeks Placebo once daily Eligibility Stage IIA IIIA Stage IB IIIA, predominantly non-squamous cell carcinoma, with/without adjuvant chemotherapy Korean trial (NCT 02795884) Intercalated: erlotinib 150 mg once daily days 8 21, pemetrexed/cisplatin day 1, four cycles every three weeks; maintenance: erlotinib 150 mg once daily 1 year NP four cycles every three weeks Stage IB IIIA non-squamous cell carcinoma Japanese WJOG6410L trial Gefitinib 250 mg once daily, 24 months NP four cycles every three weeks Stage II III Chinese EVIDENCE trial (NCT02448797) Icotinib 125 mg three times daily until PD NP four cycles every 3 weeks Stage II IIIA adenocarcinoma, no right pneumonectomy Chinese ICTAN trial (NCT 01996098) Arm A: icotinib 125 mg three times daily, 6 months; arm B: icotinib 125 mg three times daily, 12 months Platinum-based chemotherapy four cycles Stage IIA IIIA ICWIP (NCT 02125240) Icotinib 125 mg three times daily until PD Placebo three times daily until PD Stage II IIIA adenocarcinoma, adjuvant four cycles cisplatin-based chemotherapy Primary end-point DFS DFS DFS DFS DFS DFS DFS Secondary end-points OS, 3-year DFS, 2, 3 and 5-year DFS, OS, AEs OS, AEs, type of OS OS, AEs, QoL OS, AEs, HRQoL 5-year DFS and OS, AEs, QoL OS, HRQoL, pharmacokinetics recurrence Target recruitment 220 patients 700 patients 227 patients 230 patients 320 patients 477 patients 300 patients EGFR: epidermal growth factor receptor; TKI: tyrosine kinase inhibitor; PD: progressive disease; NP; cisplatin, vinorelbine; DFS: disease-free survival; OS: overall survival; AEs: adverse events; QoL: quality of life; HRQoL: health-related quality of life. THORACIC ONCOLOGY F. MCDONALD ET AL.

Adjuvant immunotherapy Another promising adjuvant systemic therapy being investigated is immunotherapy. One of these approaches is targeting MAGE-A3. MAGE-A3 is expressed in 35% of NSCLC patients. A phase II RCT showed that in patients with stage IB II resected NSCLC expressing MAGE-A3 the HR for DFS was in favour of treatment with recombinant MAGE-A3 protein in combination with an immunostimulant; however, this was not statistically significant (HR 0.75, 95% CI 0.46 1.23; p=0.248) [113]. In the phase III MAGRIT RCT, presented at ESMO 2014 13 824 patients were screened for MAGE-A3 and 2270 were randomised between adjuvant MAGE-A3 immunotherapy or placebo. Unfortunately, MAGE-A3 immunotherapy did not improve outcome (level B evidence) [114]. The full paper has not been published yet. Other interesting immunotherapy strategies currently being explored in several phase III RCTs are the use of checkpoint inhibitors. As these studies started enrolment in 2015, results are expected within a few years. An overview of these studies is provided in table 9. Adjuvant anti-angiogenic agents As angiogenesis is one of the hallmarks of cancer and important for growth and metastatic potential of tumours, the addition of bevacizumab to platinum-based chemotherapy was tested in a phase III RCT (E1505, n=1501) and presented at the World Lung Cancer Conference 2015 [93]. Patients with resected IB (>4 cm) to IIIA NSCLC were treated with four cycles of cisplatin with either vinorelbine, pemetrexed, docetaxel or gemcitabine, and randomised between addition of bevacizumab for 1 year or chemotherapy alone. At time of interim analysis (all patients randomised), no difference was observed in the primary endpoint overall survival (HR 0.99 (95% CI 0.81 1.21; p=0.93)) (level B evidence) and in the secondary endpoint DFS (HR 0.98 (95% CI 0.84 1.14; p=0.75)). As low molecular weight heparin may also improve outcome, it was tested in the NVALT-8 trial (n=202) whether the addition of nadroparin to adjuvant chemotherapy (cisplatin combined with pemetrexed (non-squamous) or gemcitabine (squamous) improved DFS in resected stage I IIIA NSCLC patients. Results were presented at the ASCO 2016 conference: no significant differences in DFS were found, which is different from the results in the abstract [115]. In summary, international guidelines recommend the use of adjuvant platinum-containing chemotherapy in patients with completely resected stage II disease and consideration of its use in patients with resected stage IB disease with a primary tumour 4 cm. According to the proposed 8th TNM edition, these are all grouped in stage II [15, 53]. Age is not a selection criterion per se, but co-existence of severe comorbidity may lead to detrimental outcome. Currently no predictive biomarkers of clinical benefit are available and new treatment strategies, such as EGFR-TKIs, angiogenesis inhibitors and immunomodulation have not yet resulted in improved outcome, but additional trials are ongoing. TABLE 9 Planned/ongoing phase III randomised controlled trials comparing adjuvant immunotherapy to placebo after complete resection in NSCLC patients IMpower010 (NCT02486718) ANVIL (NCT02595944) EORTC-ETOP PEARLS trial (NCT02504372) Canadian Cancer Trials Group (NCT02273375) Experimental arm Atezolizumab 1200 mg intravenous 16 cycles every three weeks Nivolumab intravenous every two weeks, up to 1 year Pembrolizumab 200 mg intravenous every three weeks up to 1 year MEDI4736 intravenous up to 1 year Comparator arm observation observation placebo placebo Eligibility Stage IB ( 4 cm) IIIA, adjuvant cisplatin Stage IB ( 4 cm) IIIA, adjuvant cisplatin based Stage IB ( 4 cm) IIIA, adjuvant cisplatin Stage IB ( 4 cm) IIIA, ± adjuvant chemotherapy based chemotherapy chemotherapy tested for PD-L1, EGFR/ALKbased chemotherapy tested for PD-L1 Primary end-point DFS DFS, OS DFS DFS in PD-L1+ patients Secondary end-points OS, AEs, pharmacokinetics, immunogenicity 1, 2, 5 year DFS and OS, AEs, outcomes according to PD-L1 patterns and mutational load OS, LCSS DFS in all patients, OS (PD-L1+/all), LCSS (PD-L1+/all), AEs, QoL, cost utility ratio, pharmacokinetics Target recruitment 845 patients 714 patients 1380 patients 1100 patients PD-L1: programmed death-ligand1; EGFR: epidermal growth factor receptor; ALK: anaplastic lymphoma kinase; DFS: disease-free survival; OS: overall survival; AEs: adverse events; LCSS: lung cancer specific survival. https://doi.org/10.1183/13993003.00764-2016 12